Fault and Error Containment of Gateways in Distributed Real-Time Systems

Abstract

The composition of large embedded computer systems out of heterogenous clusters requires gateways that resolve property mismatches between networks with different communication protocols and diverse computational models. At the same time, fault and error containment mechanisms are needed in order to prevent common mode failures that could be induced by the propagation of faults between clusters. This paper presents gateways to systematically tackle these challenges. The introduced gateways contain structured collections of time-sensitive variables associated with timing information (called a realtime database). The real-time database facilitates fault and error containment by temporally decoupling clusters. For accessing the real-time database from attached networks, we introduce network adaptors that ensure compliance with temporal specifications expressed using state machines with timing constraints. The containment of faults within clusters is of particular importance for mixed-criticality systems and increases the robustness of the overall systems. The proposed solution is experimentally evaluated in an example system with a time-triggered Ethernet network and a Controller Area Network (CAN).